Paper edit by API: build a rough cut from transcript lines

To build a rough cut from transcript lines by API, run POST /v1/video-inspect with transcribe: true and sentence segmentation, choose the lines you keep, then send those ranges as video[] slots to POST /v1/timeline-1.0/render. Sume does not pick the lines for you; your code or an agent does.

Adobe's Premiere 26.5 announcement lists Paper Edit as a way to "assemble rough cuts directly from your transcript". This post covers the same idea through Sume's public routes, from the Video inspect and Timeline 1.0 docs, read 2026-09-30.

How do I get transcript lines with timing?

Send transcribe: true and segmentation: { "mode": "sentence" } to POST /v1/video-inspect. The result carries text, words[] and gapless sentence segments[] shaped like caption lines. Transcription is the only billed half: $0.01 per audio minute. Add frames: false if you do not want stills.

curl -X POST https://api.sume.com/v1/video-inspect \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: paper-edit-inspect-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/interview.mp4",
    "frames": false,
    "transcribe": true,
    "segmentation": { "mode": "sentence" }
  }'

How do the chosen lines become a sequence?

Each kept line becomes one video[] slot: source_url is the interview, source_in is the line's in-point, duration is its length, and start is where it lands on the output. video[0].start must be 0 and later starts must increase. A Timeline also needs an audio spine with audio.duration_seconds (1 to 1800); to keep the interview's own sound, extract it with POST /v1/audio-detach and list the same ranges in audio.parts[] (each part takes url, source_in and duration; at most 20 parts).

What does Sume not do here?

The docs describe routes, not an editing UI, and none of the routes above chooses lines for you. Which lines to keep is a judgment that stays in your code or your agent.

Should I trim each line first or use source_in?

source_in on a slot reads a range straight out of the file, so a separate trim is not needed for a plain cut. Use video-trim only when you want each range as its own reusable MP4. Run /v1/timeline-1.0/plan before you render to see duration_seconds, segment_count and estimated_cost_usd_micros without creating a job.